MadSpin and MadGraph decay widths

Asked by Sihyun Jeon on 2018-01-24

Hi I put my cards in here.

The issue is this :
we are making
process p p > n1 mu [QCD]
with
decay n1 > ptl ptl ptl
(here, ptl are some mixtures of leptons and jets)

For other normal mass points for n1 (15~75 GeV) & (85~2000 GeV) the generations and cross section results are fine.
However, for mn1 = 80 GeV there's a bit problem.
“CRITICAL : Branching ratio larger than one for 9900012"
We get an error message like above and the result is odd.
Branching ratio for that mass point is bigger than 1 which does not make sense physically
(I put the cards in [1] and you will be able to reproduce them)

I guess that it is due to the mn1 = 80 GeV is very similar to mass of W boson 80.4 GeV and this makes the decay weird.
However we still are not sure where this problem is from, yet.

In this kind of situation, is there any suggestions to get a correct cross section for 80gev mass point?
And do you think it's ok to believe the event kinematics which madgraph produces in these kind of cases?

Thanks, Sihyun Jeon.

PS I just checked LO generation by :
(generate p p > n1 l, n1 > l j j)
and compared it with
(generate p p > n1 l)
Above is "exclusive dilepton channel" generation and below is "inclusive n1 production" so we would expect
xsec(p p > n1 l, n1 > l l j j) = xsec(p p > n1 l) * BR(n1 > l j j ).
Howver my result shows that BR(n1 > l j j) > 1.0 which does not make sense for LO as well, so this is not a problem caused by use of MadSpin...

[1] https://www.dropbox.com/sh/o6j86c63c5fecih/AACsskdH35rppFmCNHyCV_94a?dl=0

Question information

Language:
English Edit question
Status:
Solved
For:
MadGraph5_aMC@NLO Edit question
Assignee:
No assignee Edit question
Solved by:
Olivier Mattelaer
Solved:
2018-01-26
Last query:
2018-01-26
Last reply:
2018-01-26

Hi,

Let me explain how the branching ratio is computed for spinmode=none.
In this case, we simply compute it from the ratio of width.
The numerator is the width associate to your process and compute at LO by madevent. (so in your case this is the width associate to n1 > ptl ptl ptl)
and the denominator is the total width taken for your param_card.

Having that ratio larger than one means that you have a wrong value for the total width.
From your param_card it seems that you are using "Auto" for that width. Is it the case? (and is it done for that benchmark?)
If it is by how much is the BR>1? (It can also be a fluctuation and that this is a false warning)

Cheers,

Olivier

> On 24 Jan 2018, at 10:36, Sihyun Jeon <email address hidden> wrote:
>
> New question #663487 on MadGraph5_aMC@NLO:
> https://answers.launchpad.net/mg5amcnlo/+question/663487
>
> Hi I put my cards in here.
>
> The issue is this :
> we are making
> process p p > n1 mu [QCD]
> with
> decay n1 > ptl ptl ptl
> (here, ptl are some mixtures of leptons and jets)
>
> For other normal mass points for n1 (15~75 GeV) & (85~2000 GeV) the generations and cross section results are fine.
> However, for mn1 = 80 GeV there's a bit problem.
> “CRITICAL : Branching ratio larger than one for 9900012"
> We get an error message like above and the result is odd.
> Branching ratio for that mass point is bigger than 1 which does not make sense physically
> (I put the cards in [1] and you will be able to reproduce them)
>
> I guess that it is due to the mn1 = 80 GeV is very similar to mass of W boson 80.4 GeV and this makes the decay weird.
> However we still are not sure where this problem is from, yet.
>
> In this kind of situation, is there any suggestions to get a correct cross section for 80gev mass point?
> And do you think it's ok to believe the event kinematics which madgraph produces in these kind of cases?
>
>
> Thanks, Sihyun Jeon.
>
> PS I just checked LO generation by :
> (generate p p > n1 l, n1 > l j j)
> and compared it with
> (generate p p > n1 l)
> Above is "exclusive dilepton channel" generation and below is "inclusive n1 production" so we would expect
> xsec(p p > n1 l, n1 > l l j j) = xsec(p p > n1 l) * BR(n1 > l j j ).
> Howver my result shows that BR(n1 > l j j) > 1.0 which does not make sense for LO as well, so this is not a problem caused by use of MadSpin...
>
> [1] https://www.dropbox.com/sh/o6j86c63c5fecih/AACsskdH35rppFmCNHyCV_94a?dl=0
>
> --
> You received this question notification because you are an answer
> contact for MadGraph5_aMC@NLO.

Sihyun Jeon (shjeon) said : #2

It's like BR > 4.

Yes and indeed we are using auto for the width calculation.

This behaviour is only shown when mass of n1 is near 80 GeV, (below 79 and above 82 are fine)

Is there any way that you suggest us to calculate this cross section correctly or is it technically hard thing to be done in madgraph

Hi,

In your model you have the mass of the W which is:
7.995123e+01
so just below the n1 mass of 80.

In that case, you are in case where NWA can not be fully used (in this case for the W). The width computation assumes that the width
can assume that the W is onshell and here this introduces a bias.

So this is kind of expected to have issue for that benchmark point with the automatic width computation.

Cheers,

Olivier

> On 25 Jan 2018, at 20:52, Sihyun Jeon <email address hidden> wrote:
>
> Question #663487 on MadGraph5_aMC@NLO changed:
> https://answers.launchpad.net/mg5amcnlo/+question/663487
>
> Status: Answered => Open
>
> Sihyun Jeon is still having a problem:
> It's like BR > 4.
>
> Yes and indeed we are using auto for the width calculation.
>
> This behaviour is only shown when mass of n1 is near 80 GeV, (below 79
> and above 82 are fine)
>
> Is there any way that you suggest us to calculate this cross section
> correctly or is it technically hard thing to be done in madgraph
>
> --
> You received this question notification because you are an answer
> contact for MadGraph5_aMC@NLO.

Sihyun Jeon (shjeon) said : #4

Thanks Olivier Mattelaer, that solved my question.